White balance oddity: Leica or Lighroom?

[marchyman]

Advertisement (gone after registration)

I'm learning my M11. The past couple of days I've been playing with colors and white balance. I created a profile from a ColorChecker passport which I like better than the Adobe profiles.  My outdoors-in-noon-sun profile seems to be more accurate in the blues. Anyway, I also took test shots using different white balance selections using the same lens (C Biogon 35mm f/2.8), aperture, and shutter speed.  When I pulled the DNG files into Lightroom Classic I see different settings than than set in camera.

• Camera: 5500K.  Lightroom: 5900 + 32 (Temp + Tint)
• Camera: 5000K. Lightroom 5350 + 31
• Camera: Auto.  Lightroom: 5150 + 23
• Camera: Sunlight. Same as 5000K settings

Using the ColorChecker to white balance all images resulted in a proper balance of 5600 + 9.  The temperature wasn't far off from my settings and/or expected values, but the tint way too far into the magenta.  Does that come from the camera, Lightroom, or both?

[Antonio Russell]

In camera WB is irrelevant when shooting RAW. You can change it to your hearts content in post-

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[frame-it]

4 minutes ago, Antonio Russell said:

In camera WB is irrelevant when shooting RAW. You can change it to your hearts content in post-

except when shooting under orange street lights at night

[elmars]

vor einer Stunde schrieb marchyman:

  Does that come from the camera, Lightroom, or both?

It comes from the camera. There is a long thread in the German forum about this problem. Tint is about 20 point into plus, so all photos get a magenta cast. We reported this to Leica. 

[Al Brown]

Raws are easily fixable in Lightroom/ACR until a solution from this CAMERA RELATED problem shows up in (probably) next FW update from Leica.

[elmars]

Let's hope that Leica will fix it. The more of us complain at Leica the more probable this is. More in this thread (please use deepl.com to translate)

 

[SrMi]

Advertisement (gone after registration)

8 hours ago, frame-it said:

except when shooting under orange street lights at night

Why?

[marchyman]

17 hours ago, elmars said:

It comes from the camera. There is a long thread in the German forum about this problem. Tint is about 20 point into plus, so all photos get a magenta cast. We reported this to Leica. 

Thanks!  And to all the "who cares, they are raw files" people -- I care 🙂  Yes, the white balance is easily adjusted in Lr.   I still like seeing something closer to proper colors before I start editing.   My solution is to apply a preset on import that changes the white balance to 5600K + 9.  That gives me something that lets me compare and select images without being annoyed by the magenta cast.

[elmars]

vor 13 Stunden schrieb marchyman:

"who cares, they are raw files"

The JPG are wrong too. 

[Adam Bonn]

In a very OVER SIMPLIFIED way DNG/adobe WB works like this

When we shoot a RAW file a tag is embedded into it (ASN = As Shoot Neutral)

The WB (and colours) are a mathematical equation which we will MASSIVELY over simplify* here as

ASN x Profile = kelvin values and colours

If a camera has an unwanted colour bias in it’s WB algorithm (such as some are saying the M11 has) then this needs to addressed by the OEM

ASN effectively means ‘this is white, base the colour science on this being neutral’ so if it’s not (really neutral) then changing the profile part (by making your own) of 

ASN x Profile = kelvin values and colours

Will have limited success, as ASN will give it a bum steer.
………

*from memory…. The real maths is something like a mired CIEXYZ value displayed as kelvin taken from the inverse of the interpolating ColorMatrices multiplied by the ASN tag. (Actually thinking about it, there may be a XY to XYZ conversion in there be fore the kelvin value too…. Whatever, I doubt anyone cares that much)

[marchyman]

When looking at and comparing the metadata of images shit with different white balance settings I saw changes in the White Point.  Example:

     Auto white balance setting: White Point: 0.3445 0.3672

     Daylight white balance setting: White Point: 0.3374 0.3625

The other fields that I thought might have an effect on color were the same in all images:

 

  Camera Calibration 1            : 1 0 0 0 1 0 0 0 1

  Camera Calibration 2            : 1 0 0 0 1 0 0 0 1

  Calibration Illuminant 1        : Standard Light A

  Calibration Illuminant 2        : D65

It's probably obvious that I know little to nothing about this subject 🙂

[Adam Bonn]

On 3/23/2023 at 6:21 PM, marchyman said:

When looking at and comparing the metadata of images shit with different white balance settings I saw changes in the White Point.  Example:

     Auto white balance setting: White Point: 0.3445 0.3672

     Daylight white balance setting: White Point: 0.3374 0.3625

The other fields that I thought might have an effect on color were the same in all images:

 

  Camera Calibration 1            : 1 0 0 0 1 0 0 0 1

  Camera Calibration 2            : 1 0 0 0 1 0 0 0 1

  Calibration Illuminant 1        : Standard Light A

  Calibration Illuminant 2        : D65

It's probably obvious that I know little to nothing about this subject 🙂

 

In the tool you're using to poke around in the DNG you'll see something like this, this is from an M10R though, so the numbers in the other cameras will be different

 

EXIF	Color Matrix 1	0.7124023438 -0.25 -0.04345703125 -0.5737304688 1.552246094 0.3596191406 -0.1096191406 0.2922363281 0.7744140625
EXIF	Color Matrix 2	0.4877929688 -0.1286621094 -0.04858398438 -0.583984375 1.370117188 0.220703125 -0.1804199219 0.3369140625 0.5251464844

You will also see something like

EXIF

As Shot Neutral

0.2853957637 1 0.6684073107

 This is the white point of the RAW data

In a super basic easy to type, not very factually accurate way

ASN * Color Matrix = whitebalance

Adobe use a different of color matrices than Leica do (hence adobe colours and leica colours)

When you make your own profile you get a 3rd set of color matrices

But the ASN doesn't change in any one DNG file (it changes between DNG files, but there's only one per DNG and it's set when you make the photo)

So

ASN * adobe Color Matrix = whitebalance1

ASN * Leica Color Matrix = whitebalance2

ASN * home made Color Matrix = whitebalance3

The real maths is closer to this, but the principle above is sound

On 3/22/2023 at 2:48 PM, Adam Bonn said:

*from memory…. The real maths is something like a mired CIEXYZ value displayed as kelvin taken from the inverse of the interpolating ColorMatrices multiplied by the ASN tag. (Actually thinking about it, there may be a XY to XYZ conversion in there before the kelvin value too…. Whatever, I doubt anyone cares that much)

 

On 3/23/2023 at 6:21 PM, marchyman said:

Camera Calibration 1            : 1 0 0 0 1 0 0 0 1

  Camera Calibration 2            : 1 0 0 0 1 0 0 0 1

  Calibration Illuminant 1        : Standard Light A

  Calibration Illuminant 2        : D65

This is just standard profile stuff....

2 colour matrices, one for D65 one for StdA...

It looks like this

"CalibrationIlluminant1": "StdA",

"CalibrationIlluminant2": "D65",

"ColorMatrix1": [

[ 0.549800, -0.192900, -0.033500 ],

[ -0.442800, 1.198000, 0.277500 ],

[ -0.084600, 0.225500, 0.597700 ]

],

"ColorMatrix2": [

[ 0.493200, -0.130100, -0.049100 ],

[ -0.590400, 1.385200, 0.223100 ],

[ -0.182400, 0.340600, 0.530900 ]

], 

 

When you shoot the colorchecker target you get new numbers than the ones above

[milescrist]

moral of the story is that whether it's a Sony a7R (II III IV V) or leica M11 you should set the white balance in camera as closely to what you want it to ultimately be, otherwise 'fixing it in post' by changing it in Lightroom or capture one or any other raw converter will lead to strange color issues especially with certain shades. many people don't know this but raw is not really RAW with these still cameras. if you want real raw you need a RED cinema camera. Just another reason I didn't shell out ten grand for a scratchy buggy M11. Hopefully they'll fix this issue amongst the others with the M11-P, for an extra $500 of course.